Active Testing: Sample-Efficient Model Evaluation

Hi, good to see you here! 👋

This is code for "Active Testing: Sample-Efficient Model Evaluation".

Please cite our paper, if you find this helpful:

@article{kossen2021active,
  title={{A}ctive {T}esting: {S}ample-{E}fficient {M}odel {E}valuation},
  author={Kossen, Jannik and Farquhar, Sebastian and Gal, Yarin and Rainforth, Tom},
  journal={arXiv:2103.05331},
  year={2021}
}

Setup

The requirements.txt can be used to set up a python environment for this codebase. You can do this, for example, with conda:

conda create -n isactive python=3.8
conda activate isactive
pip install -r requirements.txt

Reproducing the Experiments

• To reproduce a figure of the paper, first run the appropriate experiments

  sh reproduce/experiments/figure-X.sh

• And then create the plots with the Jupyter Notebook at

  notebooks/plots_paper.ipynb

• (The notebook let's you conveniently select which plots to recreate.)

• Which should put plots into notebooks/plots/.

• In the above, replace X by

  • 123 for Figures 1, 2, 3
  • 4 for Figure 4
  • 5 for Figure 5
  • 6 for Figure 6
  • 7 for Figure 7

• Other notes

  • Synthetic data experiments do not require GPUs and should run on pretty much all recent hardware.
  • All other plots, realistically speaking, require GPUs.
  • We are also happy to share a 4 GB file with results from all experiments presented in the paper.
  • You may want to produce plots 7 and 8 for other experiment setups than the one in the paper, i.e. ones you already have computed.
  • Some experiments, e.g. those for Figures 4 or 6, may run a really long time on a single GPU. It may be good to
    • execute the scripts in the sh-files in parallel on multiple GPUs.
    • start multiple runs in parallel and then combine experiments. (See below).
    • end the runs early / decrease number of total runs (this can be very reasonable -- look at the config files in conf/paper to modify this property)
  • If you want to understand the code, below we give a good strategy for approaching it. (Also start with synthetic data experiments. They have less complex code!)

Running A Custom Experiment

• main.py is the main entry point into this code-base.

  • It executes a a total of n_runs active testing experiments for a fixed setup.
  • Each experiment:
    • Trains (or loads) one main model.
    • This model can then be evaluated with a variety of acquisition strategies.
    • Risk estimates are then computed for points/weights from all acquisition strategies for all risk estimators.

• This repository uses Hydra to manage configs.

  • Look at conf/config.yaml or one of the experiments in conf/... for default configs and hyperparameters.
  • Experiments are autologged and results saved to ./output/.

• See notebooks/eplore_experiment.ipynb for some example code on how to evaluate custom experiments.

  • The evaluations use activetesting.visualize.Visualiser which implements visualisation methods.
  • Give it a path to an experiment in output/path/to/experiment and explore the methods.
  • If you want to combine data from multiple runs, give it a list of paths.
  • I prefer to load this in Jupyter Notebooks, but hey, everybody's different.

• A guide to the code

  • main.py runs repeated experiments and orchestrates the whole shebang.
    • It iterates through all n_runs and acquisition strategies.
  • experiment.py handles a single experiment.
    • It combines the model, dataset, acquisition strategy, and risk estimators.
  • datasets.py, aquisition.py, loss.py, risk_estimators.py all contain exactly what you would expect!
  • hoover.py is a logging module.
  • models/ contains all models, scikit-learn and pyTorch.
    • In sk2torch.py we have some code that wraps torch models in a way that lets them be used as scikit-learn models from the outside.

And Finally

Thanks for stopping by!

If you find anything wrong with the code, please contact us.

We are happy to answer any questions related to the code and project.